Classification Method and System for Small Collections of High-Value Entities

1. A method for classifying and displaying small collections of high-value entities that have missing data, wherein high-value entities are measurable items whose individual worth is substantial enough to justify classification even if a total number of the high-value entities is too small to attain statistical significance, the method comprising: gathering available data about relevant attributes of high-value entities that are to be classified, wherein the relevant attributes include one dependent variable and multiple independent variables for each high-value entity, wherein the dependent variable identifies a particular high-value entity, and wherein the independent variables are values that represent answers to questions asked about the particular high-value entity; preparing the available data, about each high-value entity, for analysis by: validating values for the available data, converting non-numeric values that make up the available data into numeric values, and inverting scales as needed such that increases in each independent variable lead to increases in a correlating dependent variable; computing a computed weight for each independent variable associated with the high-value entities, wherein the computed weight is set to zero for any independent variable that has missing values, and wherein the computed weight is increased above a standard baseline weight value for any independent variable that has no missing values; scoring each high-value entity variable by multiplying each independent variable, of said each high-value entity, by the computed weight to create a score for every high-value entity; classifying all high-value entities as cases that have a similar calibration score that is based on a) multiplying each independent variable by its computed weigh to create a weight product, b) summing the weight products for all independent variables into a score for each dependent variable, and c) calculating a combination of scores for each dependent variable to classify similar cases of high-value entities; categorizing each high-value entity according to the calculated combination of scores for each dependent variable assigned to the entity; and representing all high-value entities as high-value entity representations in a calibrated visual model, wherein a newly classified high-value entity, which has missing data in its independent variables, is displayed in a graphical manner such that similarly scored high-value entities are represented in close proximity to one another.

2. The method of claim 1 , wherein the calibrated visual model is divided and categorized into multiple zones, the method further comprising: establishing a guard band between zones in the calibrated visual model, wherein high-value entities whose high-value entity representations fall in the guard band are considered to be only tentatively classified, and wherein the initial position of a high-value entity representation relative to a guard band is incorporated into the step of computing weighting factors computation of weights such that high-value entity representations are moved out of the guard band in order to improve zone classification of the high-value entities.

3. The method of claim 1 , wherein the high-value entities are selected from a group consisting of ultra-large-scale projects, unique projects, customer segments, product brands, market geographies, service types, legislation and regulations.

4. The method of claim 1 , wherein the step of preparing the available data, about each high-value entity, for analysis further comprises transforming the values to reduce any severe non-normalities that are present.

5. The method of claim 1 , wherein the step of preparing the available data, about each high-value entity, for analysis further comprises rescaling variables such that variable means and variance are approximately the same.